N-glycosylation affects substrate specificity of chicory fructan 1-exohydrolase: evidence for the presence of an inulin binding cleft


Author for correspondence: Wim Van den Ende
Tel: +32 16321952
Fax: +32 16321967
Email: wim.vandenende@bio.kuleuven.be


  • • Recently, the three-dimensional structure of chicory (Cichorium intybus) fructan 1-exohydrolase (1-FEH IIa) in complex with its preferential substrate, 1-kestose, was determined. Unfortunately, no such data could be generated with high degree of polymerization (DP) inulin, despite several soaking and cocrystallization attempts.
  • • Here, site-directed mutagenesis data are presented, supporting the presence of an inulin-binding cleft between the N- and C-terminal domains of 1-FEH IIa. In general, enzymes that are unable to degrade high DP inulins contain an N-glycosylation site probably blocking the cleft. By contrast, inulin-degrading enzymes have an open cleft configuration.
  • • An 1-FEH IIa P294N mutant, introducing an N-glycosylation site near the cleft, showed highly decreased activity against higher DP inulin. The introduction of a glycosyl chain most probably blocks the cleft and prevents inulin binding and degradation.
  • • Besides cell wall invertases, fructan 6-exohydrolases (6-FEHs) also contain a glycosyl chain most probably blocking the cleft. Removal of this glycosyl chain by site-directed mutagenesis in Arabidopsis thaliana cell wall invertase 1 and Beta vulgaris 6-FEH resulted in a strong decrease of enzymatic activities of the mutant proteins. By analogy, glycosylation of 1-FEH IIa affected overall enzyme activity. These data strongly suggest that the presence or absence of a glycosyl chain in the cleft is important for the enzyme's stability and optimal conformation.


Fructans, sucrose-based polyfructosyl chains, are, beside sucrose and starch, common reserve carbohydrates in flowering plants and bacteria (Hendry, 1993). Fructans are synthesized by an initial fructosyl transfer to one of the primary hydroxyl groups of sucrose, followed by further chain elongation (Van Laere & Van den Ende, 2002). Depending on the linkage type between the fructosyl residues, different types of fructan molecules can be distinguished (Lewis, 1993). Linear β (2,1)-fructans (inulin, principally occurring in dicots) and linear (levan) or branched (graminan) β (2,6)-fructans (occurring in bacteria and monocots) are the most common types. In plants, the degradation of fructans is catalyzed by fructan exohydrolases (FEHs) releasing one terminal fructose molecule at a time using water as acceptor. According to the linkage type they hydrolyze, FEHs can be classified into 1-FEHs (inulinases, E.C. and 6-FEHs (levanases, E.C. Different types of these FEHs have been fully characterized and cloned from fructan plants (Van den Ende et al., 2000, 2001, 2003a; Van Riet et al., 2006). Recently, the existence of 6&1-FEHs has been reported (Kawakami et al., 2005). Also, FEHs that preferentially degrade fructan trisaccharides were discovered (Benkeblia et al., 2005; Van den Ende et al., 2005). Unexpectedly, FEHs were also found in plants that do not accumulate fructans, such as Beta vulgaris and Arabidopsis thaliana (Van den Ende et al., 2003b; De Coninck et al., 2005).

In contrast to microbial FEHs that generally are able to hydrolyze sucrose as well (β-fructosidases), all plant FEHs purified so far seem to be unifunctional enzymes unable to degrade sucrose (Verhaest et al., 2007). Plant acid invertases, catalyzing the hydrolysis of sucrose, and FEHs are both hydrolases, only differing in their donor substrate specificities. The high sequence similarities between FEHs and cell wall type invertases suggest close evolutionary relationships; both enzymes might have evolved from a common ancestor (Van den Ende et al., 2000). Invertases fulfil crucial roles with respect to carbohydrate partitioning in plants (Sturm & Tang, 1999; Roitsch & Gonzalez, 2004). Similarly, FEHs are believed to fulfil important physiological functions in plants. Whenever carbon supply is needed by sink organs, FEHs can mobilize fructans by rapid hydrolysis (Morvan-Bertrand et al., 2001; Asega & Carvalho, 2004). It has also been reported that, besides their function during fructan breakdown, FEHs might be involved as β (2,1)-trimmers during graminan biosynthesis in wheat, possibly playing an important role in determining the in vivo fructan pattern (Bancal et al., 1992; Van den Ende et al., 2003a). Fructan degradation by FEH also increases the osmotic pressure, playing a key role in flower opening (Bieleski, 1993; Vergauwen et al., 2000).

Because of the close relationships found between acid invertases (vacuolar and cell wall type) and fructan-metabolizing enzymes (different types of fructosyl transferases and FEHs) at the molecular and structural levels, these enzymes are classified together within family 32 of the glycoside hydrolases (GH 32; http://www.cazy.org). Family GH 32 is combined with the related family GH 68, harboring bacterial invertases, levansucrases and inulosucrases, in the clan GH-J (Naumoff, 2001). In the last few years, several three-dimensional structures have been unraveled within this clan (Meng & Fütterer, 2003; Alberto et al., 2004; Nagem et al., 2004; Martinez-Fleites et al., 2005; Verhaest et al., 2005b, 2006). All these proteins show a common fold: they consist of an N-terminal fivefold β-propeller domain (GH 32 and GH 68) followed by a C-terminal domain formed by two β-sheets (only in GH 32). Superposition of these structures indicates that the active site is located in the β-propeller domain and is characterized by the presence of three highly conserved acidic residues (Reddy & Maley, 1990, 1996; Batista et al., 1999; Yanase et al., 2002). Although the first domain, containing the active site, determines catalytic substrate specificities of some fructosyltransferases (Altenbach et al., 2004), a possible role in substrate recognition for the second domain cannot be excluded.

The three-dimensional structure of the recombinant, Pichia pastoris-derived chicory 1-FEH IIa has been described (Verhaest et al., 2005b). Heterologous expression in P. pastoris is now recognized as a valid technique to determine the functionality of plant FEHs and invertases (De Coninck et al., 2005; Ji et al., 2007) and at least for some types of fructosyl transferases (Lasseur et al., 2006; Van den Ende et al., 2006). When heterologous and native proteins were thoroughly compared (Van den Ende et al., 2003b, 2006), it was demonstrated that those enzymes had very similar properties. The glycosyl chains attached in the recombinant proteins resemble those observed in the native enzymes (Van den Ende et al., 2006), although there is more variation in the number of extra mannoses attached to the Man3GlcNAc2 backbone (Bretthauer & Castellino, 1999).

The cavity between the two structural domains in 1-FEH IIa forms a cleft emerging from the active site (Verhaest et al., 2005b). The presence of four glycerol molecules in this cleft is indicative for sugar binding sites (Verhaest, 2005; Tremblay et al., 2006). This observation supports the idea that this cleft might represent the inulin-binding site playing an important role in the recognition of fructans of different length. Moreover, it was found that the 1-kestose bound in the active site of chicory 1-FEH IIa is positioned towards this cleft (Verhaest et al., 2007). Here, we report on site-directed mutagenesis data further supporting the presence of an inulin-binding cleft in Cichorium intybus 1-FEH IIa.

Materials and Methods

Cloning and site-directed mutagenesis

Cichorium intybus L. 1-FEH IIa, Arabidopsis thaliana cwINV1 and Beta vulgaris 6-FEH were cloned into the pPICZα A vector as described by Verhaest et al. (2004, 2005a) and Van den Ende et al. (2003b), respectively. Single amino acid substitutions were generated following the Quick Change protocol (Stratagene, La Jolla, CA, USA), using the pPICZαA-Ci1-FEH IIa/AtcwINV1/Bv6-FEH constructs as a template. For site-directed mutagenesis, the following forward oligonucleotide primers (and complementary reverse primers) were used: 5′-GTGGGCGTGGGTTAATGAAACTGATTCTC-3′ (Ci1-FEH IIa/P294N), 5′-GTGGGCGTGGGTTGATGAAACTGATTCTC-3′ (Ci1-FEH IIa/P294D), 5′-GACACGACAAGTTTCGGTGCATTTGTTG-3′ (Ci1-FEH IIa/Y457F), 5′-GGG GTTGGACTGACGAGTCATCG-3′ (AtcwINV1/N299D) and 5′-GGGGGTGGGTTGATGAATCTTTC-3′ (Bv6-FEH/N293D) (mutations are in bold). After DpnI-digestion, an additional purification step was performed (QiaQuick PCR purification Kit, Qiagen, Valencia, CA, USA), and 3 µl of the purified plasmid was used for transformation of E. coli TOP10 cells. Mutations were confirmed by sequence analysis.

Expression and purification

The methylotrophic yeast P. pastoris was used for extracellular gene expression as described in De Coninck et al. (2005). Purification of the protein from the culture supernatant was achieved by 80% ammonium sulfate precipitation. After centrifugation (40 000 g, 30 min at 4°C), the pellet was redissolved in 50 mm sodium acetate buffer, pH 5.0, also containing 0.02% (w/v) sodium azide, in order to prevent microbial growth. Subsequently, the solution was centrifuged (15 000 g, 10 min at 4°C) to spin down undissolved material. The supernatants of B. vulgaris 6-FEH wild-type and mutant enzymes were loaded on a Fast Desalting Column HR 10/10 (Amersham Biosciences, Uppsala, Sweden) for further purification. In the case of chicory 1-FEH IIa wild-type and mutant enzymes, subsequent purification steps were as described in Verhaest et al. (2004). The supernatants of the AtcwINV1 wild-type and mutant enzymes were dialyzed against 50 mm sodium acetate buffer, pH 5.0, containing 0.02% (w/v) sodium azide for 4 h at 4°C. As a final purification step, the dialyzed solution was loaded on to a Mono S column as described in Verhaest et al. (2005a). Protein concentrations were determined by the method of Sedmak & Grossberg (1977).

Deglycosylation and Q-TOF analysis on tryptic fragments

Purified protein was treated with PNGaseF as described in the protocol (Sigma-Aldrich, St Louis, MO, USA). This treatment results in the removal of the N-linked sugar residues and in the substitution of the Asn into an Asp residue. After SDS-PAGE, the proteins bands were subjected to trypsin digestion and mass spectrometric identification by Q-TOF analysis, as previously described (Van den Ende et al., 2001). Sequence information was derived from the MS/MS spectra with the aid of the maxent 3 (de-convoluting and de-isotoping of data) and pepseq software (Micromass BioLynx software package, Manchester, UK).

Enzyme activity determinations

Appropriate aliquots of enzyme were mixed with 2 mm of different substrates in 50 mm sodium acetate buffer, pH 5.0, also containing 0.02% (w/v) sodium azide. Reaction mixtures were incubated at 30°C for different time periods. Total enzyme activity was determined by measuring the amount of fructose released by anion exchange chromatography with pulsed amperometric detection (HPAEC-PAD; Dionex, Sunnyvale, CA, USA) (Van den Ende & Van Laere, 1996). For all reactions, different time points were analyzed, ranging between 15 min and 1 h depending on the enzyme concentration and substrate used. Only data from the linear range were used and experiments were repeated three times with consistent results.

Fructan substrates

Chicory fructans were obtained as follows: a boiling water extract was prepared from chicory roots harvested in September. After centrifugation, the supernatant was subsequently loaded and eluted from a Ca-Dowex column as described by Timmermans et al. (2001). Inulin (Sigma-Aldrich), 1-kestose (TCI Europe nv, Antwerp, Belgium) and sucrose are commercially available substrates; levan and 6-kestose were generous gifts from Dr Iizuka (Iizuka et al., 1993); and 1,1-nystose was prepared as described by Van den Ende et al. (2003a).

Results and discussion

Glycosylation of Cichorium intybus 1-FEH IIa

The structure of chicory 1-FEH IIa in complex with its preferential substrate 1-kestose has recently been described (Verhaest et al., 2007). Unfortunately, no structural data of 1-FEH IIa in complex with purified inulins of higher DP (> 3) are available, despite several soaking and cocrystallization attempts. A cleft (dashed line, Fig. 1), emerging from the active site (red, Fig. 1a), is observed between the N-terminal (green, Fig. 1a) and C-terminal (blue, Fig. 1a) domains of 1-FEH IIa. There are two strong indications that this cleft might represent the inulin-binding site. Firstly, the position of 1-kestose in the active site shows that its terminal glucose is pointing towards the cleft (Fig. 1b). Secondly, several glycerol molecules were found in this cleft. It is well accepted that glycerol molecules, being polyols, are good indicators for sugar-binding sites (Verhaest, 2005; Tremblay et al., 2006). Moreover, the absence of a putative N-glycosylation site (N-X-S/T) near the cleft seems to be a general feature of all 1-FEHs that are capable of degrading higher DP inulin-type fructans (Fig. 2). On the contrary, plant enzymes that are unable to hydrolyze higher DP inulins (such as cell wall invertases and 6-FEHs) contain a putative glycosylation site (Fig. 2), most probably blocking the cleft as observed in AtcwINV1 (Verhaest et al., 2006). Most likely, 6-FEHs bind their levan substrates in a different way and at a different location.

Figure 1.

(a) General overview of the three-dimensional structure of chicory fructan 1-exohydrolase (1-FEH IIa), showing the presence of a cleft (dashed line) between the N-terminal (green) and C-terminal (blue) domains. The active site is indicated in red. (b) Detail of 1-kestose (pink) bound in the active site (red) of 1-FEH IIa. Residues subjected to mutagenesis are indicated in orange. Figures are prepared with PyMol (Delano, 2002). Note the different enzyme orientation in (a), better visualizing the cleft, compared with (b).

Figure 2.

Multiple sequence alignment of the Pro294 region (using EMBL-EBI Clustal W); the presence of an N-glycosylation site is shown in bold. The following sequences with corresponding accession numbers were used for the alignment: Cichorium intybus fructan 1-exohydrolase (1-FEH IIa) (AJ295033), C. intybus 1-FEH IIb (AJ295034), C. intybus 1-FEH I (AJ242538), Campanula rapunculoides 1-FEH (AJ509808), Triticum aestivum 1-FEH w1 (AJ516025), T. aestivum 1-FEH w2 (AJ508387), T. aestivum 6&1-FEH (AB089269), Arabidopsis thaliana 6- and 1-FEH (AY060533), Beta vulgaris 6-FEH (AJ508534), A. thaliana 6-FEH (AB029310), T. aestivum 6-FEH (AM075205), Solanum tuberosum cell wall invertase 1 (Q43171), S. tuberosum cell wall invertase 2 (Q9M4K7), S. tuberosum cell wall invertase 3 (Q9M4K8), Lycopersicum esculentum cell wall invertase5 (AJ272304), Daucus carota cell wall invertase 2 (X78424), D. carota cell wall invertase 3 (X78423), D. carota cell wall invertase 1 (X69321), Vicia faba cell wall invertase (Z35162), Pisum sativum cell wall invertase (AF063246), V. faba cell wall invertase 2 (Z35163), Chenopodium rubrum cell wall invertase (X81792), A. thaliana cell wall invertase 1 (AY079422). Functionally characterized proteins are marked with an asterisk. The question mark (?) indicates that the inulin-degrading capacity is unknown (not tested).

To confirm the presence of an inulin-binding cleft in 1-FEH IIa, a site-directed mutagenesis study was performed (P294N) creating an N-glycosylation site near the cleft (Fig. 1b). Confirmation about the glycosylation of the P294N mutant protein was obtained after PNGaseF treatment. This treatment results in the removal of N-linked glycosyl chains and substitution of the Asn residue to an Asp residue. Subsequent MS Q-TOF analysis on tryptic fragments showed the presence of the peptide VLWAWVDETDSQADDIEK, confirming that the P294N mutant was indeed glycosylated. Figure 3(a) gives an overview of the specific activities of 1-FEH IIa wild-type and P294N mutant proteins towards different substrates. A general decrease in specific activity is observed for the P294N mutant when compared with the wild-type enzyme, indicating that the introduction of a glycosyl chain might interfere with enzyme folding, influencing overall conformation and catalytic efficiency. However, the data clearly show that the P294N mutant, in particular, was highly affected in its ability to hydrolyze high DP inulin when compared with the effect on the activity towards the other substrates (Fig. 3a). The inulin/1-kestose degradation ratio decreased about 10-fold (Fig. 3b), indicating that the introduced glycosyl chain especially prevents optimal binding of inulin. Further confirmation was found by incubating the P294N mutant and wild-type proteins together with a chicory root extract (containing inulins of different DP) and time-dependent visualization of the fructans by HPAEC-PAD (Fig. 4). The mutant P294N protein was able to break down smaller inulin molecules such as 1-kestose, and, to a lesser extent, 1,1-nystose, while higher DP inulins remained unaffected, even after very long incubation times (Fig. 4). These results are in accordance with the structural data (Fig. 1b), suggesting that introduction of a glycosyl chain most probably blocks the cleft and prevents inulin binding and degradation.

Figure 3.

(a) Specific enzyme activities (s.a., mol fructose mol−1 enzyme s−1); (b) overview of the inulin/1-kestose ratio of fructan 1-exohydrolase (1-FEH IIa) wild-type, P294N, P294D and Y457F mutant proteins. I, inulin; 1K, 1-kestose; N, 1,1-nystose; L, levan; 6K, 6-kestose; S, sucrose. Results are means ± SE for n = 3.

Figure 4.

Chromatographic patterns of a chicory (Cichorium intybus) root extract incubated with 500 ng purified fructan 1-exohydrolase (1-FEH IIa) wild-type and P294N mutant protein. Incubation times: 0, 4 and 24 h; incubation temperature, 30°C. G, glucose; F, fructose; S, sucrose; 1K, 1-kestose; N, 1,1-nystose.

In order to further confirm that these results on the P294N mutant were caused by the introduction of a glycosylation site near the active site, rather than caused by the removal of the Pro294 residue itself, a P294D mutant was constructed. As shown in Fig. 3, no differences in specific activity could be detected between the 1-FEH IIa wild-type and P294D mutant proteins, excluding a critical role for the Pro294 residue.

Overall, these data strongly support the presence of an inulin-binding cleft between the two domains of 1-FEH IIa. This cleft can be blocked by the introduction of a glycosyl chain, almost totally preventing the degradation of inulin. To the best of our knowledge, this is the first report demonstrating that N-glycosylation can affect enzyme substrate specificity.

Deglycosylation of Arabidopsis cell wall invertase 1 and Beta vulgaris 6-FEH

Unlike 1-FEHs, all cell wall invertases and 6-FEHs have a putative glycosylation site near the cleft (Fig. 2). The presence of an N-linked glycosylation on the Pro294 homologues of AtcwINV1 (Asn299) and Bv6-FEH (Asn293) was effectively confirmed by structure determination (Verhaest et al., 2006) and Q-TOF analysis (Van den Ende et al., 2003b). Interestingly, both enzymes are able to degrade 1-kestose to a limited extent but they cannot hydrolyze inulin (Van den Ende et al., 2003b; De Coninck et al., 2005), which is in accordance with the above-mentioned ‘blocked cleft’ hypothesis.

As shown in Fig. 5, the cleft present between the two domains in AtcwINV1 is blocked by the N-linked glycosylation, especially by the first N-acetyl-glucosamine. Since both invertase and 6-FEH are unable to degrade inulin, it was investigated whether this incapability is the result of the presence of a glycosyl chain in the cleft. Site-directed mutagenesis studies were performed on AtcwINV1 (N299D) and Bv6-FEH (N293D) in order to investigate whether deglycosylation affects the inulin-degrading capacity. However, no differences could be detected between the wild-type (glycosylated) and mutant (deglycosylated) proteins (data not shown). These negative results on AtcwINV1 and Bv6-FEH strongly suggest that the absence of a glycosyl chain, although probably necessary to create space for positioning inulin in the cleft, is not sufficient for efficient binding and catalysis. Most likely, 1-FEHs from fructan plants contain specific amino acids in the cleft for optimal inulin binding and stabilization.

Figure 5.

General overview of the three-dimensional structure of Arabidopsis cell wall invertase 1 showing the cleft between the N-terminal (green) and C-terminal (blue) domains blocked by the presence of a glycosyl chain (yellow) linked to Asn299. Active site residues are indicated in red. This figure was prepared with PyMol (Delano, 2002).

Both deglycosylated mutant AtcwINV1 and Bv6-FEH proteins were characterized by a strong decrease in overall specific activity (Fig. 6). Taken together with the earlier described results on 1-FEH IIa, it is clear that the introduction or removal of a glycosyl chain in the cleft between the two domains of the protein has an effect on protein folding and/or stability. The results on AtcwINV1 indicate that the presence of a glycosyl chain between the two domains might be necessary to keep cell wall invertases in a stable conformation. The stabilizing effect of glycosyl chains is well known and has been ascribed to the presence of hydrogen-bondings and hydrophobic interactions between the oligosaccharide chain and the protein (Wyss & Wagner, 1996). The effect of deglycosylation on the activity of Bv6-FEH towards 6-kestose (DP3) and high DP levan (Fig. 6b) is even more drastic as compared with AtcwINV1 (Fig. 6a). This might indicate that the glycosyl chain between the two domains of Bv6-FEH fulfils another function and might interfere with proper protein folding besides the effect on the overall stability of the protein. The effects of N-glycosylation on the folding and structure of plant proteins are well documented (for a review, see Ceriotti et al., 1998).

Figure 6.

Specific enzyme activities (s.a., mol fructose mol−1 enzyme s−1) of AtcwINV 1 wild-type and N299D mutant proteins incubated with 5 mm sucrose and 1-kestose (a), and Beta vulgaris fructan 6-exohydrolase (6-FEH) wild-type and N293D mutant proteins incubated with 5 mm levan and 6-kestose (b). (a) Closed bars, sucrose; open bars, 1-kestose; (b) closed bars, levan; open bars, 6-kestose. Results are means ± SE for n = 3.

Since such a glycosyl chain is absent in 1-FEHs, it can be hypothesized that the binding of the substrate (inulin) itself might enhance the stability of the protein and the efficiency of the catalysis. Substrate-induced conformational changes were already described for several types of enzymes, such as glycosyltransferases (Qasba et al., 2005). Alternatively, a putative strong interaction between the N-terminal (Gln272) and the C-terminal (Tyr457) domains can be observed in 1-FEH IIa (Fig. 1b) but not in AtcwINV1. To investigate whether this interaction is critical to reach a stable conformation in the absence of a glycosyl chain in the cleft, site-directed mutagenesis was performed (Y457F). However, no differences in specific activities could be observed in comparison with the wild-type protein (Fig. 3), showing that Y457 is not an essential amino acid in this respect.


Despite several soaking and cocrystallization experiments, no structural data of chicory 1-FEH IIa in complex with higher-DP inulin could be obtained. In order to investigate the binding of inulin in 1-FEHs, site-directed mutagenesis experiments were performed to introduce a glycosyl chain near the cleft localized between the N- and C-terminal domains. The data strongly suggest that this cleft is the effective inulin-binding site. Since besides cell wall invertases, 6-FEHs are also characterized by the general presence of a glycosyl chain, most probably blocking the cleft (as observed in AtcwINV1), it can be concluded that inulin binding in 1-FEHs, on the one hand, and levan binding in 6-FEHs, on the other, occur in different ways and at a different location. Deglycosylation near the cleft causes a strong decrease in overall activity of AtcwINV1 and B. vulgaris 6-FEH, assuming that the glycosylation between the two domains might fulfil a critical role in the stability and/or folding of cell wall invertases and 6-FEHs. Taken together, it can be postulated that an open cleft configuration is necessary for optimal binding of inulin and that the presence or absence of a glycosyl chain between the two domains of invertases, 6-FEHs and 1-FEHs is important with regard to the enzyme's stability and optimal conformation.


The authors thank Rudy Vergauwen for his technical assistance. WVdE and AR are supported by a grant from the Fund for Scientific Research (FSR Flanders).